An important set of French and British archives on the historical seismicity of the northern Lesser Antilles is analyzed. We present the most significant data, which help to constrain the 8 February 1843, the 16 May 1851 and the 29 April 1897 earthquakes. The 1843 event is the largest reported historical earthquake of the Lesser Antilles subduction arc, and produced intensity IX on a segment about 100 km long, between Antigua and Grande-Terre (Guadeloupe), which contradicts the 300 km long proposed by other authors, who mainly rely on Robson's catalog (1964). This suggests a magnitude 7.5 to 8.0. The absence of tsunami and noticeable vertical deformation of the coasts in the source area makes very unlikely a normal fault mechanism, but does not rule out the possibility of a subduction event, because of the low-dip angle of the interplate plane. The dominance of moderate instrumental thrust earthquakes in the source area suggests a subduction mechanism for the 1843 event. The northern and southern limits of the 1843 source area roughly coincide with the Barracuda ridge and the Tiburón ridge, respectively, which may have acted as barriers during the rupture process. The northwestern portion of the Tiburon ridge is associated with a lack of instrumental seismicity, near Grande-Terre, which could indicate a possible magnitude 7.5 to 8.0 event in the next future. Moreover, its mechanism could be normal, which would provide the additional risk of a damaging tsunami. The evaluation of seismic hazard must also take into account possible shallow, moderate earthquakes, as the 1851 and the 1897 events, which produced intensities VII to VIII in Guadeloupe, and may be related to observed major Plioquaternary faults in this island.